RNA molecules are important cellular components involved in many fundamental biological roles. Because the structural features of RNA are intimately connected to their biological function, there is great interest in predicting RNA structure from sequence. The proposed research addresses both RNA structure prediction - development of a hierarchical modeling approach using graph theory and statistical tools, and RNA design - engineering fluorescent riboswitches that can sense transcription termination, offering researches visualization of the process in vivo. The proposed approach requires development of new mathematical and statistical/computer science tools for sampling RNA graph objects efficiently in 3D - to provide an initial, coarse level of sampling - and for developing systematically structure-based statistical approaches to score RNA conformations using data-mining tools. Extensive analysis of RNA junction motifs based on known structures will be used to develop separate local and global statistical potentials to predict both local geometric orientations, as well as global long-range interactions. These RNA models and associated potentials will be combined and carefully tested in three inter-related stages of folding in the following hierarchical design: (1) Explore RNA's structural conformation space coarsely using MC sampling on tree graphs that represent RNA 2D structures embedded in 3D lattices with statistical potentials that select RNA-like conformations; (2) Improve prediction accuracy using a coarse-grained three-bead-per-nucleotide model with a higher-level statistical potential to guide 3D assembly; and (3) Refine the best candidates using dynamics simulations on full atomistic models with force-field potentials. After careful analysis followed by development and testing of the new mathematical tools, each stage of the plan will be refined to finally integrate the components. New design applications for fluorescent riboswitches will be pursued, by marrying known elements of fluorescent aptamers with ligand-based configurational rearrangements of riboswitches that control gene expression.